A SLIDING GOERTZEL ALGORITHM

This paper presents a sliding Goertzel algorithm to accurately estimat e the Fourier coefficients of multifrequency (MF) sinusoidal signals b uried in noise. The algorithm is based on second-order digital resonat ors that are tuned at the desired frequencies. The proposed method pro vides the following advantages when compared with the conventional Goe rtzel algorithm. Firstly, it computes Fourier coefficients in less tha n one signal period. Therefore, faster detection time is achieved, par ticularly when the greatest common divisor (GCD) of the input frequenc ies is small. Secondly, it is less prone to numerical overflow problem s in fixed-point arithmetic implementation. Thirdly, the algorithm is guile suitable for time varying sinusoidal signal estimation. An analy sis is undertaken to provide additional insight into the issue of requ ired acquisition time versus the desired accuracy for the proposed alg orithm. Extensive simulation tests are also included to demonstrate it s performance.